It's not really a hybrid cloud
You'll note that I don't call Scale's offering a hybrid cloud. They really want me to, and they themselves are pitching it as a hybrid cloud, but I can't call it that. As far as I am concerned, clouds have things that Scale does not. A self-service portal is one such missing feature. A robust and well-documented RESTful API is another.
Scale also lacks any of the tools required to build composable workloads. Clouds are all about building composable workloads, and then automating them with scripting. This is not a hybrid cloud.
While lacking some "cloud-like" niceties, Scale's hybrid infrastructure still does place it in some pretty rare company. Very few vendors are able to offer a service provider the ability to create a hybrid infrastructure. Fewer still offer transparent layer 2 bridging.
VMware can do this. You can order up service-provider provisioned VMware infrastructure in various flavours right now, including a hybrid infrastructure product. Of course, this being VMware and anything related to the word "cloud", such a thing is spectacularly difficult. Not to mention outrageously expensive.
When talking about the possibility of service provider clouds, however, it isn't Scale versus any of the big brand names that comes to mind. Microsoft, VMware and so forth pay little more than lip service to enabling service providers to offer hybrid infrastructure, let alone full-blown hybrid clouds. Ultimately, these behemoths want to own the customer relationship and as much of the associated revenue as possible, something that the service providers would ideally like to avoid.
As such, I see Scale competing more directly with smaller players such as Hypergrid, Yottabyte, or any of the OpenStack-in-a-can vendors like Stratoscale.
From a technical standpoint, it doesn't really matter how the service provider might choose to stand up the Scale virtual node VMs. Although, considering that Scale has not yet decided when, if or how they'll license their virtual nodes to service providers, their commercial decisions in that regard could close doors.
If Scale don't take the Apple route of insisting that virtual Scale nodes only be run on Scale appliances (which, for service providers, I sincerely doubt they'd make a requirement) then they can offer service providers a means to test the waters with minimal risk. Service providers could start by running their first few virtual Scale instances inside VMs on whatever infrastructure they already have (most likely VMware) until they proved that there was demand. Once demand is proven they could start investing in a KVM infrastructure to dedicated to standing up Scale virtual nodes.
One is an engineered, purpose-built hybrid infrastructure. The other is a clever hack that uses virtualization to make a previously existing solution do something it was never originally designed to do. It's hard not see the dichotomy between the two as some sort of metaphor for the entire IT industry.
Getting users on board
Because each customer's virtual Scale node lives inside a virtual machine, it's a little bit easier for the service provider to explain to customers how their data and workloads are isolated from other customers who also use the same shared physical infrastructure.
On the other hand, if the customer wishes to create sub-virtual data centres using the Scale model, they would have to instantiate a virtual Scale node as a VM running on top of their already-virtual Scale node. This is something that, at present, Scale has no plans of supporting. You don't get too many layers deep before this gets really complicated and overhead starts to become an issue.
The smallest businesses are unlikely to want to carve up their hybrid infrastructure into virtual data centres. For the next few years, at least, they are most likely to be interested in self-service as a means for systems administrators to instantiate infrastructure and services that will then be managed by those systems administrators.
Today's medium-sized companies, however, are eager to have individual departments managing their own IT, and are already struggling with shadow IT problems when departments turn to the public cloud to do exactly that without IT's permission. Administrators from these organisations would most likely benefit from being able to carve up infrastructure into virtual data centres, and to allow department heads to carve up their allocations by project, service, customer or other billable identifier.
All of this still applies if you take the service providers out of the equation. Google's GCP is really just a great big service provider, and Scale's current announcement is giving them a means by which to reach out to small businesses.
At some point, however, the Google/Scale offering will compete against Microsoft's Azure Stack, an engineered, purpose-built hybrid cloud connected to a competing public cloud provider. With the small difference that Microsoft's marketing and sales departments are a few orders of magnitude larger than anything a regional service provider partnership could bring to bear.
The Google-Scale announcement is interesting, both for what it is and what it could one day become. Scale will face some stiff competition. ®